Memory. 73 



tion of the second reaction and that point which marked the complete 

 exposure of the word. By this method no account is taken of Ri or its 

 position relative to the exposed words. If at the time of the second 

 exposure the subject can speak the word before any part of it appears, 

 he would have a score of 250; otherwise he would get credit for just 

 that part of the word which is not exposed at the time of the second 

 reaction. This does not of course eliminate the factor of residual 

 memory, but on the other hand it does not oppose this factor to that 

 of the " saving" which we desire to measure. If residual memory was 

 present at the time of the first reaction so as to shorten or even to make 

 it occur before the exposure of the word was complete, the second 

 reaction should be still earlier than otherwise would be expected. 



The data treated in this manner are presented in table 12. Refer- 

 ring to the column of averages at the right in section i, we see clearly 

 that the alcohol values are larger, particularly on July 4. The average 

 for normal days is 74, for alcohol days 94, with a difference in favor of 

 the alcohol values of 20. The complete exposure of a word requires 250 

 of our units. On normal days, therefore, the second reaction occurred 

 on the average at a time when the word had been |r§ exposed, as against 

 250 for alcohol, that is, approximately 70 and 60 per cent. The mem- 

 ory score, with the exception of the first normal day, June 29, is larger 

 for the first period ; the differences are therefore nearly all preceded by 

 the minus sign. They average —30 for the normal days and —45 for 

 the alcohol experiments. The results for the first normal day play a 

 particularly prominent role in sections n and in of the table. On the 

 first day the score was especially low in the first period, being only 41. 

 All the differences have the plus sign on this day, with the result that 

 the average differences and per cent differences show a minus effect of 

 alcohol for periods 2, 3, and 4, 1 and a total average difference of —15, 

 or a per cent difference of —12.4. Since the value for period 1, June 

 29, is so decidedly out of line with the first-period values for the other 

 days, while the scores for periods 2, 3, and 4 on June 29 are practically 

 what would be expected judging from the other normal days, July 1 and 

 3, it is not thought just to trust the method of differences here, but 

 rather to consider the simple difference between averages. The aver- 

 age 63 for the normal period 2 is deducted from the average of 88 for 

 the alcohol period 2, resulting in a difference of +25 (see section in). 

 The differences between averages for the other periods are +15, +14, 

 +20, and +48, with an average for all periods following the dose of 

 + 20. The probable correctness figures apply to these differences and 

 are in every case of such size as to give them a good degree of reliability. 



The memory data, presented in whatever way we chose, shows that 

 with this particular memory test and in the particular form used there 

 is a better performance following the alcohol. This confirms the re- 



1 It should be noted that there were only 4 periods on June 29. 



